Analysis And Modeling Of Urban Mobility Based On Big Data

With the development of urbanization and motorization,the distribution of population and travel demand are changing rapidly.While the traditional residents' travel acquisition method,that is,the Household Travel Survey is difficult to adapt to the changing due to the long cycle and low sample size,which is a challenge for urban traffic management and control.In order to fully understand the characteristics of urban traffic,more and more cities need the intelligent traffic information collection method to obtain urban travel characteristic data at low cost and high frequency.At the same time,with the advancement of Information Communications Technology(ICT),it's easy to obtain large sample,multi-dimensional and fine-grained information.However,it is difficult to achieve the effective use of massive information because of lack of multi-dimensional data analysis and mining methods.This paper employs mobile phone signaling data and automatic license plate identification data to extract resident travel information,establish a tensor decomposition model to identify the multi-dimensional travel information,reveal the Spatial and temporal pattern of travel,and propose the bus service evaluation index considering residents' travel demand.The thesise proposes a method to identify the travel origin-destination(OD)based on trajectory data(mobile phone signaling data and automatic license plate identification data).Firstly,the applicability of data is analysed based on data quality and data coverage,then the OD recognition process method is proposed.After that,the statistical analysis of the recognition results is carried out comparing with the statistical yearbook data and resident travel survey data to verify the rationality of the data.The multidimensional data analysis model,tensor decomposition model,is proposed to analyze the spatiotemporal pattern and spatiotemporal correlation of multidimensional travel data.Firstly,the non-negative tensor decomposition model,a core tensor size determination method based on AIC criteria and the non-negative matrix decomposition initialization strategy are proposed to improve the original model.Secondly,the proposed model is applied to the trips OD tensor and motor vehicle OD tensor,and then the factor matrix of each dimension and core tensor are analyzed.The results show that the tensor decomposition model effectively identifies the morning and evening peak patterns and the non-peak pattern of the day,as well as the weekday pattern and weekend pattern of the week in the temporal mode.In the spatial mode,the patterns divide the city into different groups.The results of Pearson correlation coefficient show that pattern value of the core zone of the group has a strong positive correlation with the population of the zone.Combined with hierarchical clustering and k-means clustering,the zones are clustered according to the mean value and coefficient of variation of the pattern values.The results show that the spatial relationship between the zone and other zones can be caught.To assess whether the public transit service is well accessible for trips of specific origins,destinations and origin-destination(OD)pairs,a novel measure,the Transit Coverage Index(TCI),is proposed.TCI takes into account both the trip coverage of transit systems and the spatial distribution of heterogeneous and dynamic individual travel demands.An easy-to-implement method is also developed in this paper to extract the information of transit services and driving routes for millions of requests using the computer programming and Baidu Map.The results of the case study suggest that the TCI better represents the trip coverage of the transit system and provides a more powerful assessment tool of transit quality of service.Finally,the sensitivity of the index to parameters such as transfer times and walking distance thresholds are analyzed.Based on the automatic license plate recognition data,the model of the number of vehicles in the road network is proposed,and the verification analysis is carried out based on the measured data.However,the data-based model belongs to the off-line analysis after the event,and the real-time calculation cannot be realized.The paper combines the two-flow theory to derive the model of the number of vehicles in the road network and combines the simulation method to verify the model and provide support for traffic demand management.